Take the 2-minute tour ×
Sound Design Stack Exchange is a question and answer site for sound engineers, producers, editors, and enthusiasts. It's 100% free, no registration required.

Why clipped sound perceived like someone is pouring sand on drum's membrane? It feels like it happens on specific frequencies and not totally random. Which maybe means it can be filtered?

Can we filter clipping distortion based on its specific properties? (i'm not talking about spline interpolation but more like cut-off approach maybe)

Added

Just wanted to clarify that I'm aware of the nature of clipping phenomena but more concerned about specifics of perception of this phenomena. So, after DFT/FFT of the clipped signal we're having those "nasty frequencies". Are they sharing any properties in common? are they located within some specific frequency range? Can we simply cut them off? Can we cut them off in specific cases, like when the spectral characteristics of clipped signal are well known and predictable?

The whole point of asking this is that i want to cleanup this record http://www.2shared.com/audio/SMuu991E/smrad.html and alike. It's MP3 and I don't have wav version. I feel like that 'sandy' noise has some predictable nature and identifiable by my ear. It's not like abstract smoothing of the signal, and it's clearly lies within some range, which i believe should be distinguishable from the useful signal. Please help if you know what i'm talking about.

That story is tricky because my record is in MP3 form, meaning that i have spectral form only. I have no way of getting original wav.

share|improve this question

migrated from avp.stackexchange.com Jan 24 at 12:01

This question came from our site for engineers, producers, editors, and enthusiasts spanning the fields of video, and media creation.

2 Answers 2

The question about the nature of clipping phenomena was discussed many times here on this site (for example here) and on dozens of other places which are easily googled.

So as it was already discussed, clipping - is just few points where signal amplitude exceed the available range of finite amplitude levels. This causing the wave signal to become less continuous which makes its spectral representation much more complex.

Imagine a single sine tone which contain only one frequency. In the spectral domain such signal will be represented as a single frequency. However when the signal become clipped (i.e. not smooth and continuous), its spectral representation become much more complex and many additional frequencies are introduced. This is easily visible in the below image:

enter image description here

This is the reason why all those nasty frequencies (pouring sand on drum's membrane) appear.

Theoretically, it is possible to attempt to repair such signal using interpolation[extrapolation]4 techniques which will make the sound wave more "round" in the clipped areas, however it may or may not produce any audible results. More on clipping repair here.

share|improve this answer
    
thanks for the answer, just wanted to clarify that i'm aware of "the nature of clipping phenomena" but more concerned about specifics of perception of this phenomena. So, after DFT/FFT we're having those "nasty frequencies". Are they sharing any properties in common? are they located within some specific frequency range? Can we simply cut them off? Can we cut them off in specific cases, like when the spectral characteristics of clipped signal are well known and predictable? –  ILICH Apr 2 '13 at 16:55
    
When the signal is getting clipped, some of the information is getting lost. You cannot filter out the frequencies which were created as a result of this loss. These frequencies only there to describe a more complex soundwave. If you will try to filter out more frequencies it will cause even larger losses. The only way to handle clipping (not necessarily very successful) is interpolating it in its time domain. Nothing more you can do. –  Eugene S Apr 2 '13 at 17:29
    
Let's make one more note: <b>some</b> information is getting lost. And I'm ok with that actually. What bothers me is the fact that additional information aka 'sandy noice' was introduced. I want to remove that information because it's not useful in my specific case and caused by FFT transformation of the encoding mechanism (i suppose, correct me if i'm wrong here, cause i may be) –  ILICH Apr 2 '13 at 17:34

Clipping occurs when the level of signal needed to accurately render a sound exceeds the ability of the system in question to handle the signal. This can happen at several different points, ranging from the signal level that can be produced by the amplifier to the distance a speaker can travel.

Ultimately, the result is a partially square wave which results in a jerky motion on play back as a speaker can not render a square wave effectively due to momentum. The sound you hear is a product of the speaker failing to accurately reproduce the square wave. The way around it is to remove the square wave by doing something like spline interpolation.

share|improve this answer
    
thank you for the answer, i forgot to mention that the point of asking this question is to figure out how to fix some specific record and i don't have wav, but only mp3. It means that it's already in spectral representation and i can't use interpolation effectively –  ILICH Apr 2 '13 at 17:14

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.